366 Jeffrey and Cole —Experimental Investigations 
the scalariform perforations obliterated by the reduction in calibre and 
development to realize structures of the same nature as those occurring 
traumatically in Drimys . This interpretation, moreover, gains force from 
the fact that in Liriodendron such degenerate vessels are of normal, although 
rare, occurrence. Further, the general phenomena of traumatism lead us to 
expect, more often than not, the recall of ancestral characters in an abnormal 
form. This is, for example, pre-eminently true of the traumatic resin 
canals and traumatic ray tracheides of the Conifers. A final argument for 
the interpretation of the curious elements appearing in the root of Drimys 
as a consequence of injury as reversionary indications of the former presence 
of vessels in the genus is that such an hypothesis explains the fact satis- 
factorily. If the opposition and fusion of pits in the tracheides of the root 
in Drimys were only a meaningless vagary, we should expect to find 
parallel conditions in the injured woods of Conifers. Such have never been 
described. 
We may now pass to the question of the inherent probability of the 
suppression of vessels in angiospermous woods. The evidence appears 
to be overwhelmingly in favour of such a possibility. Fig. 12, PI. VII, 
illustrates part of the woody cylinder of Alnus japonica . In the centre 
there is a broad radial zone of the wood devoid of vessels, and laterally 
several similar non-vascular stripes of less diameter may be seen. The 
broad central band of vessel-less wood corresponds in position to the leaf 
trace. Such conditions are found in a number of cases in woody Dicotyle- 
dons, and are of very wide occurrence among the herbaceous representatives 
of the group. It has been shown by investigations carried out by students 
of this laboratory that the evascularization of wood is a phenomenon com- 
monly related to the transformation of regions of the woody cylinder into 
parenchyma. In the case of Alnus figured above, it is quite easy to observe 
the gradual blotting out of the typical vascular organization as one 
progresses from the region of the leaf gap outwards. Since there is 
absolutely no question that vessels may degenerate in the Dicotyledons 
locally, there seems to be no difficulty in regarding as possible the occur- 
rence of this phenomenon as a general feature of organization, particularly 
as this situation is actually realized in certain Cactaceae and Crassulaceae, as 
indicated above. Further, if the Gnetales or similar forms are ancestral to 
or cognate with the Angiosperms, the possession of vessels is clearly a 
primitive characteristic of the higher seed plants known as Angiosperms. 
In conclusion we may attempt to picture to ourselves the type of wood 
structure from which Drimys has been derived. As has been pointed out in 
the beginning, the general organization of the wood, apart from the absence 
of vessels, is that of one of our northern oaks. There is good reason, both 
on account of its early abundant occurrence as a fossil and likewise on com- 
parative anatomical grounds, to regard the oak as a relatively primitive 
